jam13 - 11:07 am on Oct 24, 2004 (gmt 0)

I'd skip the completely passive cooling method. Chips can warm up pretty hot these days, and you don't want to wreak havoc on your system, or have your processor burn out prematurely.

Whilst I admit that with the ever increasing power consumption of the faster CPUs, passive cooling _is_ becoming impractical, I still think that passive component cooling with good case airflow (ducting) is preferable to the heatsink + fan per component approach in most situations. Clone PCs have far too many fans in these days due to:

1) Cost - a small heatsink + fan is cheaper than a good passive heatsink.

2) Smaller PC assemblers and home builders don't want to worry about designing a system with good thermal performance - it's easier to just cool each component in isolation and hope for the best.

The problems with this approach are that you end up with a noisy system because of all the little fans, and you also introduce several potential mechanical points of failure. Big manufacturers have been using passive cooling even on the CPU for years in servers and workstations because it's more reliable (bigger/slower fans last longer) and more efficient. The reduction in noise seems to be something that has only become important more recently.

Unfortunately it's hard to get hold of passive CPU heatsinks - most of the available ones are low profile designs for 1u racks. And you can't just take the fan off a normal cooler - the fins on the heatsink are designed for forced cooling (fast airflow) and won't work without it.

The old 'ATX' case / mobo design has been good for 30 years.

30 years?!